Merge remote-tracking branch 'upstream/master' into deprecate-nonkeyword-args-clip

Author: MarcoGorelli

doc/source/whatsnew/v1.3.0.rst

@@ -648,7 +648,9 @@ Deprecations
 - Deprecated setting :attr:`Categorical._codes`, create a new :class:`Categorical` with the desired codes instead (:issue:`40606`)
 - Deprecated behavior of :meth:`DatetimeIndex.union` with mixed timezones; in a future version both will be cast to UTC instead of object dtype (:issue:`39328`)
 - Deprecated using ``usecols`` with out of bounds indices for ``read_csv`` with ``engine="c"`` (:issue:`25623`)
-- Deprecated passing arguments (other than ``"lower"`` and ``"upper"``) as positional in :meth:`DataFrame.clip` and :meth:`Series.clip` (:issue:`41485`)
+- Deprecated passing arguments as positional in (:issue:`41485`) :
+  - :meth:`DataFrame.clip` and :meth:`Series.clip` (other than ``"lower"`` and ``"upper"``)
+  - :meth:`DataFrame.interpolate` (except for ``"method"``) and :meth:`Series.interpolate`
 
 .. ---------------------------------------------------------------------------
 

pandas/core/arrays/datetimelike.py

@@ -599,7 +599,9 @@ def _validate_shift_value(self, fill_value):
                 "will raise in a future version, pass "
                 f"{self._scalar_type.__name__} instead.",
                 FutureWarning,
-                stacklevel=8,
+                # There is no way to hard-code the level since this might be
+                #  reached directly or called from the Index or Block method
+                stacklevel=find_stack_level(),
             )
             fill_value = new_fill
 

pandas/core/arrays/datetimes.py

@@ -1175,6 +1175,7 @@ def to_perioddelta(self, freq) -> TimedeltaArray:
             "future version.  "
             "Use `dtindex - dtindex.to_period(freq).to_timestamp()` instead",
             FutureWarning,
+            # stacklevel chosen to be correct for when called from DatetimeIndex
             stacklevel=3,
         )
         from pandas.core.arrays.timedeltas import TimedeltaArray

pandas/core/frame.py

@@ -1764,6 +1764,7 @@ def to_dict(self, orient: str = "dict", into=dict):
                 "will be used in a future version. Use one of the above "
                 "to silence this warning.",
                 FutureWarning,
+                stacklevel=2,
             )
 
             if orient.startswith("d"):
@@ -10647,6 +10648,29 @@ def clip(
     ) -> DataFrame | None:
         return super().clip(lower, upper, axis, inplace, *args, **kwargs)
 
+    @deprecate_nonkeyword_arguments(version=None, allowed_args=["self", "method"])
+    def interpolate(
+        self: DataFrame,
+        method: str = "linear",
+        axis: Axis = 0,
+        limit: int | None = None,
+        inplace: bool = False,
+        limit_direction: str | None = None,
+        limit_area: str | None = None,
+        downcast: str | None = None,
+        **kwargs,
+    ) -> DataFrame | None:
+        return super().interpolate(
+            method,
+            axis,
+            limit,
+            inplace,
+            limit_direction,
+            limit_area,
+            downcast,
+            **kwargs,
+        )
+
 
 DataFrame._add_numeric_operations()
 

pandas/core/generic.py

@@ -481,13 +481,19 @@ def _data(self):
     @property
     def _AXIS_NUMBERS(self) -> dict[str, int]:
         """.. deprecated:: 1.1.0"""
-        warnings.warn("_AXIS_NUMBERS has been deprecated.", FutureWarning, stacklevel=3)
+        level = self.ndim + 1
+        warnings.warn(
+            "_AXIS_NUMBERS has been deprecated.", FutureWarning, stacklevel=level
+        )
         return {"index": 0}
 
     @property
     def _AXIS_NAMES(self) -> dict[int, str]:
         """.. deprecated:: 1.1.0"""
-        warnings.warn("_AXIS_NAMES has been deprecated.", FutureWarning, stacklevel=3)
+        level = self.ndim + 1
+        warnings.warn(
+            "_AXIS_NAMES has been deprecated.", FutureWarning, stacklevel=level
+        )
         return {0: "index"}
 
     @final
@@ -6696,7 +6702,6 @@ def replace(
         else:
             return result.__finalize__(self, method="replace")
 
-    @final
     def interpolate(
         self: FrameOrSeries,
         method: str = "linear",

pandas/core/indexes/base.py

@@ -3691,7 +3691,7 @@ def is_int(v):
                     "and will raise TypeError in a future version.  "
                     "Use .loc with labels or .iloc with positions instead.",
                     FutureWarning,
-                    stacklevel=6,
+                    stacklevel=5,
                 )
             indexer = key
         else:

pandas/core/indexes/datetimes.py

@@ -40,6 +40,7 @@
     cache_readonly,
     doc,
 )
+from pandas.util._exceptions import find_stack_level
 
 from pandas.core.dtypes.common import (
     DT64NS_DTYPE,
@@ -660,7 +661,7 @@ def _deprecate_mismatched_indexing(self, key) -> None:
                     "raise KeyError in a future version.  "
                     "Use a timezone-aware object instead."
                 )
-            warnings.warn(msg, FutureWarning, stacklevel=5)
+            warnings.warn(msg, FutureWarning, stacklevel=find_stack_level())
 
     def get_loc(self, key, method=None, tolerance=None):
         """

pandas/core/reshape/merge.py

@@ -673,6 +673,8 @@ def __init__(
                 f"in a future version. ({left.columns.nlevels} levels on the left,"
                 f"{right.columns.nlevels} on the right)"
             )
+            # stacklevel chosen to be correct when this is reached via pd.merge
+            # (and not DataFrame.join)
             warnings.warn(msg, FutureWarning, stacklevel=3)
 
         self._validate_specification()

pandas/core/series.py

@@ -5271,6 +5271,29 @@ def clip(
     ) -> Series | None:
         return super().clip(lower, upper, axis, inplace, *args, **kwargs)
 
+    @deprecate_nonkeyword_arguments(version=None, allowed_args=["self", "method"])
+    def interpolate(
+        self: Series,
+        method: str = "linear",
+        axis: Axis = 0,
+        limit: int | None = None,
+        inplace: bool = False,
+        limit_direction: str | None = None,
+        limit_area: str | None = None,
+        downcast: str | None = None,
+        **kwargs,
+    ) -> Series | None:
+        return super().interpolate(
+            method,
+            axis,
+            limit,
+            inplace,
+            limit_direction,
+            limit_area,
+            downcast,
+            **kwargs,
+        )
+
     # ----------------------------------------------------------------------
     # Add index
     _AXIS_ORDERS = ["index"]

pandas/tests/arrays/test_datetimelike.py

@@ -561,7 +561,8 @@ def test_shift_fill_int_deprecated(self):
         data = np.arange(10, dtype="i8") * 24 * 3600 * 10 ** 9
         arr = self.array_cls(data, freq="D")
 
-        with tm.assert_produces_warning(FutureWarning, check_stacklevel=False):
+        msg = "Passing <class 'int'> to shift"
+        with tm.assert_produces_warning(FutureWarning, match=msg):
             result = arr.shift(1, fill_value=1)
 
         expected = arr.copy()
@@ -783,10 +784,13 @@ def test_to_perioddelta(self, datetime_index, freqstr):
         dti = datetime_index
         arr = DatetimeArray(dti)
 
-        with tm.assert_produces_warning(FutureWarning):
+        msg = "to_perioddelta is deprecated and will be removed"
+        with tm.assert_produces_warning(FutureWarning, match=msg):
             # Deprecation GH#34853
             expected = dti.to_perioddelta(freq=freqstr)
-        with tm.assert_produces_warning(FutureWarning, check_stacklevel=False):
+        with tm.assert_produces_warning(
+            FutureWarning, match=msg, check_stacklevel=False
+        ):
             # stacklevel is chosen to be "correct" for DatetimeIndex, not
             #  DatetimeArray
             result = arr.to_perioddelta(freq=freqstr)

pandas/tests/dtypes/cast/test_promote.py

@@ -406,11 +406,13 @@ def test_maybe_promote_any_with_datetime64(
         exp_val_for_scalar = fill_value
 
     warn = None
+    msg = "Using a `date` object for fill_value"
     if type(fill_value) is datetime.date and dtype.kind == "M":
         # Casting date to dt64 is deprecated
         warn = FutureWarning
 
-    with tm.assert_produces_warning(warn, check_stacklevel=False):
+    with tm.assert_produces_warning(warn, match=msg, check_stacklevel=False):
+        # stacklevel is chosen to make sense when called from higher-level functions
         _check_promote(dtype, fill_value, expected_dtype, exp_val_for_scalar)
 
 

pandas/tests/dtypes/test_missing.py

@@ -443,8 +443,10 @@ def test_array_equivalent(dtype_equal):
 )
 def test_array_equivalent_series(val):
     arr = np.array([1, 2])
+    msg = "elementwise comparison failed"
     cm = (
-        tm.assert_produces_warning(FutureWarning, check_stacklevel=False)
+        # stacklevel is chosen to make sense when called from .equals
+        tm.assert_produces_warning(FutureWarning, match=msg, check_stacklevel=False)
         if isinstance(val, str)
         else nullcontext()
     )

pandas/tests/frame/methods/test_diff.py

@@ -285,3 +285,12 @@ def test_diff_readonly(self):
         result = df.diff()
         expected = DataFrame(np.array(df)).diff()
         tm.assert_frame_equal(result, expected)
+
+    def test_diff_all_int_dtype(self, any_int_dtype):
+        # GH 14773
+        df = DataFrame(range(5))
+        df = df.astype(any_int_dtype)
+        result = df.diff()
+        expected_dtype = "float32" if any_int_dtype in ("int8", "int16") else "float64"
+        expected = DataFrame([np.nan, 1.0, 1.0, 1.0, 1.0], dtype=expected_dtype)
+        tm.assert_frame_equal(result, expected)

pandas/tests/frame/methods/test_interpolate.py

@@ -342,3 +342,15 @@ def test_interp_fillna_methods(self, axis, method):
         expected = df.fillna(axis=axis, method=method)
         result = df.interpolate(method=method, axis=axis)
         tm.assert_frame_equal(result, expected)
+
+    def test_interpolate_pos_args_deprecation(self):
+        # https://github.com/pandas-dev/pandas/issues/41485
+        df = DataFrame({"a": [1, 2, 3]})
+        msg = (
+            r"In a future version of pandas all arguments of DataFrame.interpolate "
+            r"except for the argument 'method' will be keyword-only"
+        )
+        with tm.assert_produces_warning(FutureWarning, match=msg):
+            result = df.interpolate("pad", 0)
+        expected = DataFrame({"a": [1, 2, 3]})
+        tm.assert_frame_equal(result, expected)

pandas/tests/frame/methods/test_join.py

@@ -345,7 +345,11 @@ def test_merge_join_different_levels(self):
         # join, see discussion in GH#12219
         columns = ["a", "b", ("a", ""), ("c", "c1")]
         expected = DataFrame(columns=columns, data=[[1, 11, 0, 44], [0, 22, 1, 33]])
-        with tm.assert_produces_warning(FutureWarning, check_stacklevel=False):
+        msg = "merging between different levels is deprecated"
+        with tm.assert_produces_warning(
+            FutureWarning, match=msg, check_stacklevel=False
+        ):
+            # stacklevel is chosen to be correct for pd.merge, not DataFrame.join
             result = df1.join(df2, on="a")
         tm.assert_frame_equal(result, expected)
 

pandas/tests/frame/methods/test_quantile.py

@@ -4,6 +4,7 @@
 import pandas as pd
 from pandas import (
     DataFrame,
+    Index,
     Series,
     Timestamp,
 )
@@ -650,3 +651,68 @@ def test_quantile_ea_scalar(self, index, frame_or_series):
             assert result == expected
         else:
             tm.assert_series_equal(result, expected)
+
+    @pytest.mark.parametrize(
+        "dtype, expected_data, expected_index, axis",
+        [
+            ["float64", [], [], 1],
+            ["int64", [], [], 1],
+            ["float64", [np.nan, np.nan], ["a", "b"], 0],
+            ["int64", [np.nan, np.nan], ["a", "b"], 0],
+        ],
+    )
+    def test_empty_numeric(self, dtype, expected_data, expected_index, axis):
+        # GH 14564
+        df = DataFrame(columns=["a", "b"], dtype=dtype)
+        result = df.quantile(0.5, axis=axis)
+        expected = Series(
+            expected_data, name=0.5, index=Index(expected_index), dtype="float64"
+        )
+        tm.assert_series_equal(result, expected)
+
+    @pytest.mark.parametrize(
+        "dtype, expected_data, expected_index, axis, expected_dtype",
+        [
+            pytest.param(
+                "datetime64[ns]",
+                [],
+                [],
+                1,
+                "datetime64[ns]",
+                marks=pytest.mark.xfail(reason="#GH 41544"),
+            ),
+            ["datetime64[ns]", [pd.NaT, pd.NaT], ["a", "b"], 0, "datetime64[ns]"],
+        ],
+    )
+    def test_empty_datelike(
+        self, dtype, expected_data, expected_index, axis, expected_dtype
+    ):
+        # GH 14564
+        df = DataFrame(columns=["a", "b"], dtype=dtype)
+        result = df.quantile(0.5, axis=axis, numeric_only=False)
+        expected = Series(
+            expected_data, name=0.5, index=Index(expected_index), dtype=expected_dtype
+        )
+        tm.assert_series_equal(result, expected)
+
+    @pytest.mark.parametrize(
+        "expected_data, expected_index, axis",
+        [
+            [[np.nan, np.nan], range(2), 1],
+            [[], [], 0],
+        ],
+    )
+    def test_datelike_numeric_only(self, expected_data, expected_index, axis):
+        # GH 14564
+        df = DataFrame(
+            {
+                "a": pd.to_datetime(["2010", "2011"]),
+                "b": [0, 5],
+                "c": pd.to_datetime(["2011", "2012"]),
+            }
+        )
+        result = df[["a", "c"]].quantile(0.5, axis=axis)
+        expected = Series(
+            expected_data, name=0.5, index=Index(expected_index), dtype=np.float64
+        )
+        tm.assert_series_equal(result, expected)

pandas/tests/frame/methods/test_reset_index.py

@@ -341,7 +341,7 @@ def test_reset_index_with_datetimeindex_cols(self, name):
         )
         df.index.name = name
 
-        with tm.assert_produces_warning(warn, check_stacklevel=False):
+        with tm.assert_produces_warning(warn):
             result = df.reset_index()
 
         item = name if name is not None else "index"

pandas/tests/frame/methods/test_to_dict.py

@@ -81,7 +81,8 @@ def test_to_dict_invalid_orient(self):
     def test_to_dict_short_orient_warns(self, orient):
         # GH#32515
         df = DataFrame({"A": [0, 1]})
-        with tm.assert_produces_warning(FutureWarning, check_stacklevel=False):
+        msg = "Using short name for 'orient' is deprecated"
+        with tm.assert_produces_warning(FutureWarning, match=msg):
             df.to_dict(orient=orient)
 
     @pytest.mark.parametrize("mapping", [dict, defaultdict(list), OrderedDict])

pandas/tests/frame/test_constructors.py

@@ -2681,3 +2681,14 @@ def test_from_out_of_bounds_timedelta(self, constructor, cls):
         result = constructor(scalar)
 
         assert type(get1(result)) is cls
+
+    def test_nested_list_columns(self):
+        # GH 14467
+        result = DataFrame(
+            [[1, 2, 3], [4, 5, 6]], columns=[["A", "A", "A"], ["a", "b", "c"]]
+        )
+        expected = DataFrame(
+            [[1, 2, 3], [4, 5, 6]],
+            columns=MultiIndex.from_tuples([("A", "a"), ("A", "b"), ("A", "c")]),
+        )
+        tm.assert_frame_equal(result, expected)